On Wednesday, April 14, 2004, at 12:20 PM, Dar Scott wrote:
Seconds are seconds.
My attempt at brevity (I yak too much and I'm feeling crumby today, anyway) might make my comments terse, cryptic and even abrasive. I'll try again.
The seconds are uniform and continuous and well defined. (Except... see below.)
If I had anything serious to do with dates and times, I'd first look at Sarah Reicheit's library stack. I have not seen it, but it sounds promising. I would either build on that, make my own functions or, should speed be a problem, find ways to make sure the built-in commands do what I want. I would consider those in that order.
If you use seconds for sub-second work, then you shouldn't use convert or time during that on some systems; they interact. However, if you use your own date-time functions based on seconds, it is OK. I assume so on all systems.
If the clock on the computer is corrected, the seconds is adjusted. Though this is a problem for those using sockets or send or timing things, this is normally what is needed when thinking of date and time. Changing to/from daylight savings is not a correction in the sense I'm using here; it is a view.
Perhaps, this way, by using your own functions and seconds you avoid the sub-second timing problems, the use of system time-zone in conversions, the alleged problems with daylight savings, and the zone problems in Internet time conversions.
Dar Scott
_______________________________________________ use-revolution mailing list [EMAIL PROTECTED] http://lists.runrev.com/mailman/listinfo/use-revolution
